Swarming behaviour of salps algorithm for predicting chemical compound activities

Abstract

The increasing size of chemical search space of chemical compound databases and importance of similarity measurements to drug discovery are main factors in chem.-informatics research. This paper introduces a swarming behavior of salps algorithm for predicting chemical compound activities. The salp optimization algorithm is proposed for chemical descriptor selection with three initialization (small, mixed and large). The K-nearest neighbor (KNN) was utilized for the fitness function of salps swarm optimization algorithm (SSOA) to choose a small number of features and achieve high classification accuracy. Experimental results reveal the capability of SSA to find an optimal feature subset which maximizes the classification performance and minimizes the number of selected features. A set of assessment indicators are used to evaluate and compared with different algorithms inclduing particle swarm optimization (PSO), Grasshopper Optimization Algorithm (GOA), Grey Wolf Optimizer(GWO), Sine Cosine Algorithm (SCA), Whale Optimization algorithm (WOA) using three initialization method and a superior accuracy was obtained with our proposed approach. Also, in comparison with other algorithms that used the same data, our approach has a higher performance using less number of features. The previous algorithms (GOA, GWO, PSO, SSA, SCA, WOA) are compared and three different methods are used to initialize the different optimization algorithms to ensure capability of the different optimizers to converge from different initial positions namely mixed initialization, small initialization, and large initialization.